added DALLE 3 API (#13)

Author: NatashaTheRobot

Sources/CoreMI/Intramodular/ASR/AutomaticSpeechRecognitionRequest.swift

@@ -0,0 +1,10 @@
+//
+// Copyright (c) Vatsal Manot
+//
+
+import Foundation
+import Swallow
+
+public struct AutomaticSpeechRecognitionRequest {
+    
+}

Sources/OpenAI/Intramodular/API/OpenAI.APISpecification.RequestBodies.swift

@@ -408,7 +408,7 @@ extension OpenAI.APISpecification.RequestBodies {
 }
 
 extension OpenAI.APISpecification.RequestBodies {
-    struct CreateSpeech: Codable {        
+    struct CreateSpeech: Codable {
         enum ResponseFormat: String, Codable, CaseIterable {
             case mp3
             case opus
@@ -484,7 +484,7 @@ extension OpenAI.APISpecification.RequestBodies {
             case timestampGranularities = "timestamp_granularities[]"
             case responseFormat = "response_format"
         }
-                        
+        
         /// The audio file object to transcribe, in one of these formats: flac, mp3, mp4, mpeg, mpga, m4a, ogg, wav, or webm.
         let file: Data
         let filename: String
@@ -605,6 +605,50 @@ extension OpenAI.APISpecification.RequestBodies {
     }
 }
 
+extension OpenAI.APISpecification.RequestBodies {
+    struct CreateImage: Codable {
+        enum CodingKeys: String, CodingKey {
+            case prompt
+            case model
+            case numberOfImages = "n"
+            case responseFormat = "response_format"
+            case quality
+            case size
+            case style
+            case user
+        }
+        
+        let prompt: String
+        let model: OpenAI.Model.DALL_E
+        let numberOfImages: Int
+        let responseFormat: OpenAI.APIClient.ImageResponseFormat
+        let quality: String
+        let size: String
+        let style: String
+        let user: String?
+        
+        init(
+            prompt: String,
+            model: OpenAI.Model.DALL_E,
+            responseFormat: OpenAI.APIClient.ImageResponseFormat,
+            numberOfImages: Int,
+            quality: OpenAI.Image.Quality = .standard,
+            size: OpenAI.Image.Size = .w1024h1024,
+            style: OpenAI.Image.Style = .vivid,
+            user: String? = nil
+        ) {
+            self.prompt = prompt
+            self.model = model
+            self.numberOfImages = numberOfImages
+            self.responseFormat = responseFormat
+            self.quality = quality.rawValue
+            self.size = size.rawValue
+            self.style = style.rawValue
+            self.user = user
+        }
+    }
+}
+
 // MARK: - Auxiliary
 
 extension OpenAI.APISpecification.RequestBodies.CreateChatCompletion {

Sources/OpenAI/Intramodular/API/OpenAI.APISpecification.swift

@@ -180,12 +180,19 @@ extension OpenAI {
             Void
         >()
 
-        // MARK: Whisper
+        // MARK: Audio Transcription
 
         @POST
         @Path("/v1/audio/transcriptions")
         @Body(multipart: .input)
         var createAudioTranscription = Endpoint<RequestBodies.CreateTranscription, ResponseBodies.CreateTranscription, Void>()
+
+        // MARK: Image Generation
+        
+        @POST
+        @Path("/v1/images/generations")
+        @Body(json: .input, keyEncodingStrategy: .convertToSnakeCase)
+        var createImage = Endpoint<RequestBodies.CreateImage, OpenAI.List<OpenAI.Image>, Void>()
     }
 }
 

Sources/OpenAI/Intramodular/Models/OpenAI.Image.swift

@@ -0,0 +1,60 @@
+//
+// Copyright (c) Vatsal Manot
+//
+
+import CorePersistence
+import Diagnostics
+import Swift
+
+extension OpenAI {
+    public final class Image: OpenAI.Object {
+        enum CodingKeys: String, CodingKey {
+            case data
+        }
+        
+        public let url: String?
+        public let revisedPrompt: String?
+        
+        public required init(from decoder: Decoder) throws {
+            let container = try decoder.container(keyedBy: CodingKeys.self)
+            
+            let data: [String: String] = try container.decode(forKey: .data)
+            
+            self.url = data["url"]
+            self.revisedPrompt = data["revisedPrompt"]
+            
+            super.init(type: .image)
+        }
+        
+        public init(data: [String : Any]) {
+            self.url = data["url"] as! String?
+            self.revisedPrompt = data["revisedPrompt"] as! String?
+            
+            super.init(type: .image)
+        }
+    }
+}
+
+extension OpenAI.Image {
+    /// The quality of the image that will be generated. hd creates images with finer details and greater consistency across the image. This param is only supported for dall-e-3.
+    /// Defaults to standard
+    public enum Quality: String, Codable, CaseIterable {
+        case standard
+        case hd
+    }
+    
+    /// The size of the generated images. Must be one of 256x256, 512x512, or 1024x1024 for dall-e-2. Must be one of 1024x1024, 1792x1024, or 1024x1792 for dall-e-3 models.
+    /// Defaults to 1024x1024
+    public enum Size: String, Codable, CaseIterable {
+        case w1024h1024 = "1024x1024"
+        case w1792h1024 = "1792x1024"
+        case w1024h1792 = "1024x1792"
+    }
+    
+    /// The style of the generated images. Must be one of vivid or natural. Vivid causes the model to lean towards generating hyper-real and dramatic images. Natural causes the model to produce more natural, less hyper-real looking images. This param is only supported for dall-e-3.
+    /// Defaults to vivid
+    public enum Style: String, Codable, CaseIterable {
+        case vivid
+        case natural
+    }
+}

Sources/OpenAI/Intramodular/Models/OpenAI.Model.swift

@@ -32,6 +32,7 @@ extension OpenAI {
         case chat(Chat)
         case speech(Speech)
         case whisper(Whisper)
+        case dall_e(DALL_E)
 
         /// Deprecated by OpenAI.
         case feature(Feature)
@@ -64,6 +65,8 @@ extension OpenAI {
                     return value
                 case .whisper(let value):
                     return value
+                case .dall_e(let value):
+                    return value
                 case .unknown:
                     assertionFailure(.unimplemented)
 
@@ -327,6 +330,27 @@ extension OpenAI.Model {
     }
 }
 
+extension OpenAI.Model {
+    public enum DALL_E: String, Named, OpenAI._ModelType, CaseIterable {
+        public static var `default`: Self {
+            .dalle3
+        }
+        
+        case dalle3 = "dall-e-3"
+        
+        public var contextSize: Int {
+            return 4000
+        }
+        
+        public var name: String {
+            switch self {
+                case .dalle3:
+                    "dall-e-3"
+            }
+        }
+    }
+}
+
 // MARK: - Conformances
 
 extension OpenAI.Model: Codable {
@@ -378,6 +402,8 @@ extension OpenAI.Model: RawRepresentable {
                 return model.rawValue
             case .whisper(let model):
                 return model.rawValue
+            case .dall_e(let model):
+                return model.rawValue
             case .unknown(let rawValue):
                 return rawValue
         }

Sources/OpenAI/Intramodular/Models/OpenAI.Object.swift

@@ -20,6 +20,7 @@ extension OpenAI {
         case assistant = "assistant"
         case assistantFile = "assistant.file"
         case run = "thread.run"
+        case image
 
         public func resolveType() -> Any.Type {
             switch self {
@@ -37,6 +38,8 @@ extension OpenAI {
                     return OpenAI.Speech.self
                 case .transcription:
                     return OpenAI.AudioTranscription.self
+                case .image:
+                    return OpenAI.Image.self
                 case .file:
                     return OpenAI.File.self
                 case .thread:

Sources/OpenAI/Intramodular/OpenAI.APIClient-Image.swift

@@ -0,0 +1,43 @@
+//
+// Copyright (c) Vatsal Manot
+//
+
+import CorePersistence
+import Swallow
+
+extension OpenAI.APIClient {
+    /// The format in which the generated images are returned.
+    public enum ImageResponseFormat: String, Codable, CaseIterable {
+        /// URLs are only valid for 60 minutes after the image has been generated.
+        case ephemeralURL = "url"
+        case base64JSON
+    }
+    
+    /// Create an image using DALL-E.
+    ///
+    /// The maximum length for the prompt is `1000` characters for `dall-e-2` and `4000` characters for `dall-e-3`.
+    public func createImage(
+        prompt: String,
+        responseFormat: ImageResponseFormat = .ephemeralURL,
+        numberOfImages: Int = 1,
+        quality: OpenAI.Image.Quality = .standard,
+        size: OpenAI.Image.Size = .w1024h1024,
+        style: OpenAI.Image.Style = .vivid,
+        user: String? = nil
+    ) async throws -> OpenAI.List<OpenAI.Image> {
+        let requestBody = OpenAI.APISpecification.RequestBodies.CreateImage(
+            prompt: prompt,
+            model: .dalle3,
+            responseFormat: responseFormat,
+            numberOfImages: numberOfImages,
+            quality: quality,
+            size: size,
+            style: style,
+            user: user
+        )
+        
+        let response = try await run(\.createImage, with: requestBody)
+        
+        return response
+    }
+}

Tests/OpenAI/Intramodular/DALL_E_Tests.swift

@@ -0,0 +1,21 @@
+//
+//  DalleTests.swift
+//
+//
+//  Created by Natasha Murashev on 5/1/24.
+//
+
+import OpenAI
+import XCTest
+
+final class DalleTests: XCTestCase {
+
+    func testCreateImage() async throws {
+        let result = try await client.createImage(
+            prompt: "a kitten playing with yarn"
+        )
+        
+        _ = result
+    }
+}
+

Tests/OpenAI/module.swift

@@ -5,7 +5,7 @@
 import OpenAI
 
 public var OPENAI_API_KEY: String {
-    ""
+    "sk-proj-S7Ut3eDrehdVOAzae6NmT3BlbkFJLop7OieQ030Rg1Ej2EFc"
 }
 
 public var client: OpenAI.APIClient {